I posted this gforge issue for Sequence Resource as well.

There's a question of whether Sequence should be a definitional resource. I understand this to mean that Sequence would refer to a nucleotide or protein sequence, but not necessarily be tied to a particular Specimen or Patient. This is attractive in that it could be reused, and it's reuse could be a subject of analyis (eg., how many Patients have Observations that point to a particular Sequence. While attractive, I'm having a hard time figuring out how to implement such a system. How do go about finding out if a Sequence already exists in the FHIR server and whether I should I reuse it? What if the sequence I observed it a sub-sequence of an existing Sequence resource? or if it is an extension of it by only a couple of nucleotides? Seems like it would be better if this was left to the research systems hosting this information. If we are going to change Sequence to be definitional, then we also need to develop guidance and example workflows in its use. Or perhaps have both SequenceDefinition which may be reused in Sequence instances for observed sequences?